Plant disease detection technology based on hybrid convolutional attention mechanism

As the scale of agricultural production continues to expand, plant diseases have become a major threat to global agricultural productivity. Although traditional disease detection methods are widely used, they generally suffer from low accuracy and complex operations, making them difficult to meet the needs of modern agriculture. In response to these problems, this article proposes a new deep learning model-LabInt. LabInt uses dual-branch independent convolution in Lab color space, which effectively reduces the impact of illumination changes on feature extraction. Through spatial reconstruction and multi-scale pooling technology, the model can deeply mine shallow disease features and use deep convolution instead of ordinary convolution to reduce computational costs and improve the model's fine-grained identification capabilities. In addition, in order to adapt to complex agricultural environments, LabInt introduces a multi-head self-attention (MHA) mechanism to dynamically enhance attention to disease characteristics and significantly reduce the interference of complex backgrounds. Experimental results show that LabInt exhibits excellent disease recognition performance in actual agricultural production environments, with key indicators such as precision and recall reaching high levels. Ablation studies show that each module makes an important contribution to model performance. Therefore, LabInt is not only innovative in theory, but also shows strong application potential in practice, and is expected to become an important tool to improve the efficiency of agricultural disease prevention and control.